We propose to study the social transmission of (1) conduct problems and (2) substance use across 3 generations (G1, G2, and G3), and developmental pathways of these behaviors from ages 22 months to 11 years. A dynamic developmental-contextual approach will be used to examine prospectively the transmission of contextual risk and problematic behaviors across 3 generations, and parenting behaviors across 2 generations. Hypotheses will be tested on a community sample of the children of at-risk early adult men (G2) in the Oregon Youth Study (OYS). Up to 2 biological children of the fathers per monther are included in the 3GS study. The OYS men show high levels of antisocial behavior (at age 25 years, 42% had 2 or more arrests) and substance use (estimated 50% lifetime diagnoses for alcohol dependence or abuse). The men are currently 26-28 years old, and with their parents (G1) have been involved in the OYS since the G2 men were 9-10 years old. Main hypotheses re: the development of conduct problems are based on the coercion model of family interaction and social influence, including peer deviancy contexts. Child risk factors to be examined will include temperament. We expect that initiation of substance use will be predicted by factors related to the development of conduct problems, and by the effects of exposure to substance use by parents in childhood and by peers in later childhood. Two secondary specific aims are to examine (1) the emergence of co-occurring externalizing and internalizing behaviors in childhood, and the association of these with early onset of substance use; (2) the mechanisms involved in the intergenerational transmission of risk for sexually transmitted diseases (STDs), the human immunodeficiency virus (HIV), and AIDS, by examining precursors of adolescent sexual relationships. We will compare prediction and models developed from the core theoretical approach to alternative developmental models. In particular, we will examine whether insecure or coercive attachment, cognitive deficits, and low sociocognitive skill levels or biases can account for additional variance in the development of conduct prolems, deviant peer association, and early substance use. We plan to address the specific aims by a combination of secondary data analysis of the OYS and 3GS extant data sets, from the beginning of the proposed study period, and by collecting and analyzing data that will (1) expand the sample size for the early waves (1 through 6, ages 22 months to 9 years) and (2) extend the study to age 11 years.